Three Essays in Cluster Robust Machine Learning and High-Dimensional Econometrics

Abstract

The new generation machine learning and high-dimensional techniques have become powerful tools for economists. In economics, researchers are often facing cross-sectional dependence. However, the existing methods are often established under an independent sampling assumption. Failure of accounting for such dependence can potentially lead to false positive research results. This dissertation attempts to provide a first look at some new machine learning and high-dimensional methods under various of cross-sectional dependence assumptions.